Conflicts in the Work–Family Interface: Links to Job Stress,Customer Service Employee Performance,and Customer Purchase Intent

Abstract

Because customer service employees often represent the sole contact a customer has with a firm, it is important to examine job-related factors that affect customer service employee performance and customer evaluations. In two diverse customer settings, the authors capture matched responses from service employees, supervisors, and customers. The authors use the data to examine the potential chain of effects from customer service employee work–family conflict and family–work conflict, to job stress and job performance, to customer purchase intent (CPI). The results show direct (and indirect) effects of work–family conflict and family–work conflict on service employee customer-directed extra-role performance (CDERP). The results also show direct effects of job stress on service employee in-role performance (IRP) and CDERP and on CPI. Furthermore, the findings show that job stress has a more pronounced effect on IRP than on CDERP and that CDERP has a greater effect on CPI than does IRP. The authors conclude with a discussion of managerial and theoretical implications.

In recent years, the changing workforce (e.g., more dual career couples) and the nature of work itself (e.g., an increase in the number of service sector jobs) have given rise to conflicts in the work–family interface that may affect the work performance of employees and organizations (De Jonge and Dormann 2003). The conflicts may have their most pronounced effects in the stressful environment of customer service jobs. Customer service employees may take their jobs home with them, creating work–family conflicts (WFCs) that lead to additional stress at work, thus affecting their performance and customers’ evaluations. Indeed, recent reports suggest that conflicts in the work– family interface lead to job stress, which in turn seriously impairs service employee performance (Molpus 2003; National Institute for Occupational Safety and Health 2002). Two stressors in this interface are WFC and family– work conflict (FWC); the former is a form of interrole conflict in which the demands created by the job interfere with the performance of family-related responsibilities, and the latter is a form of interrole conflict in which the demands created by the family interfere with the performance of work-related responsibilities (Netemeyer, Boles, and McMurrian 1996). As with most WFC and FWC conceptualizations, our concepts view family as a larger, nonwork entity that includes responsibilities toward spouses, children, nonmarried partners, and home life in general (Adams, King, and King 1996; Frone 2000; Frone, Russell, and Cooper 1992).

To date, there is primarily anecdotal evidence of the effects of WFC and FWC on employee performance and customer outcomes. For example, USA Today reports that 32% of employees indicated that balancing work and family demands was their leading job-related concern (Armour 2002). Likewise, a Financial Times article notes that despite the best managerial support, the home demands of customer service employees may permeate their work lives and negatively affect their job performance (Furnham 2002). Given that customer service employees are highly stressed and struggle to balance the demands of several parties, it is important to examine the potential chain of effects from WFC and FWC, to job stress, to performance, and, ultimately, to customer purchase intent (CPI). Furthermore, because service employees often need to go beyond their in-role performance (IRP) to help customers, research must explore how the work–family interface affects customer-directed extra-role performance (CDERP) and customer evaluations.

Our research examines three questions pertaining to the potential chain of effects from WFC and FWC to CPI. First, how do WFC and FWC affect customer service employee performances? Are their effects mediated by job stress, or are they direct? Second, what is the effect of job stress on customer purchase intent? Third, what are the relative effects of WFC, FWC, and job stress on service employee performances? We collect multisource data in two settings (one business-to-consumer and one business-to-business) to address these questions.

Overview and Hypotheses

Overview of Models

Figure 1 presents two models of the relationships among WFC, FWC, job stress, service employee IRP, service employee CDERP, and CPI. Although many frameworks have been proposed for studying the work–family interface, our models are derived from three mostly complementary theoretical perspectives: (1) interrole conflict theory (Kahn and Byosiere 1992), (2) identity theory (Thoits 1991), and (3) conservation of resources (COR)/resource drain theory (Hobfoll 2002). The three theories share the common premise that WFC, FWC, and job stress compete for or tax the employee resources of time, cognitive effort, and emotional and psychological energy.

Figure 1

Mediated and Incremental Effects Models of WFC and FWC

The mediated model in Figure 1, Panel A, posits that the effects of WFC and FWC on CDERP are mediated by job stress, and the effect of job stress on CPI is mediated by the two performance constructs (i.e., IRP and CDERP). This model is consistent with the more traditional approaches of how job stressors (i.e., WFC and FWC) and job stress affect performance and customer outcomes (Beehr 2000; Kahn and Byosiere 1992); that is, job stress is an affective mediator of the effects of WFC and FWC on performance, and performance is a behavioral mediator of the effect of job stress on CPI. The incremental effects model in Figure 1, Panel B, posits that FWC and WFC directly affect CDERP beyond their effects on job stress, and job stress directly affects CPI beyond the effects of the performance variables. We compare the mediated and incremental effects models with the premise that the incremental direct effects of WFC and FWC on CDERP and of job stress on CPI are tenable.

Mediated Model

Job stress and the effects of WFC and FWC

We define job stress as nervousness/anxiety associated with the job, affecting an employee's emotional and/or physical health (Cox, Griffiths, and Rial-Gonzalez 2000). Interrole conflict theory contends that WFC should affect job stress because of the competing demands that WFC places on time, cognitive, and emotional resources that are necessary to fulfill both work and family roles (Frone, Russell, and Cooper 1992). Similarly, identity theory contends that people have multiple role identities and devote resources to maintain identities that are salient to them. People experience job stress when they encounter impediments to these identities. Thus, WFC represents an impediment to successfully meeting the family role, which in turn increases job stress (Thoits 1995). In addition, the relationship between WFC and job stress is consistent with the COR/resource drain approach, which specifies that energy, cognitive effort, and so forth, are finite resources that people apply to both work and family. Thus, WFC leads to job stress because resources are lost in the process of juggling both work and family, creating elevated levels of stress at work (Hobfoll 2002).

Interrole conflict theory suggests that job stress is experienced if a person is struggling to meet the demands of a role because of interferences from the family role. As such, those who perceive high FWC may be overwhelmed by their home duties and experience job stress (Frone, Russell, and Cooper 1992). Consistent with identity theory and the COR/resource drain approach, FWC may represent an impediment to completing work duties, because the family role is competing for resources that are necessary to fulfill the work role. This undermines a person's work–role identity, increasing job stress. Scholars suggest that such an effect represents a spillover in which the demands of one role exacerbate the stress in another role (Edwards and Rothbard 2000).

H₁:

Both (a) WFC and (b) FWC positively affect (produce) job stress.

Service employees and job performance

Before offering rationale for the effects of job stress on customer service employee performances, we provide an overview of service employees’ customer-linking behaviors, and we distinguish between IRP and CDERP. At least three customer-linking behaviors of service employees have been identified (Bettencourt and Brown 2003). First, service employees can enhance their firm's image by advocating the firm's products and services to third parties (i.e., customers). Second, service employees are in a position to unveil customer wants, needs, and expectations that can be communicated to the firm to serve customers better. Third, customer evaluations are contingent on the service employees with whom the customer interacts. Such high-quality interaction enhances a firm's customer evaluations. These customer-linking behaviors can help retain current customers and attract new customers through favorable word of mouth from current customers. Embedded in these behaviors are IRP and CDERP.

Our conceptualization of IRP is consistent with three works. First, Motowidlo and Van Scotter (1994) suggest that IRP reflects officially required performances that serve the goals of or support the technical core of the firm. Second, Singh, Verbeke, and Rhoads (1996) operationalize IRP in terms of performance with regard to knowledge of the firm's products, knowledge of customer needs, management of time, company resources, and expenses, as well as the number (quantity) of customers served. Third, Singh (2000) views IRP in terms of productivity and quality. Productivity reflects quantifiable output in terms of customer contacts/ calls and backroom functions (e.g., paperwork), and quality reflects formally mandated behaviors during the employee– customer interface. Consistent with these views, we define (and operationalize) IRP as (1) the quality of performance with regard to the employee's knowledge of the company, competitor products, and customers; (2) the quality of performance with regard to the accurate management of records, time, and expenses; and (3) the quantity of work achieved.

We focus on the most relevant type of extra-role performance in the customer service setting, CDERP. Borman and Motowidlo (1993) suggest that favorably representing the organization to customers is part of “contextual” customer performance that goes beyond an employee's in-role job requirements. Bettencourt, Gwinner, and Meuter (2001) and Bettencourt and Brown (1997) suggest that CDERP reflects employees’ extra efforts to take initiatives that improve service when they communicate with customers, as well as conscientious extra efforts to respond to customer concerns. Consistent with these views, we define CDERP as the degree to which the service employee “goes the extra mile” in serving customers during the employee–customer interface.

Effects of job stress on performance

It is widely accepted that the customer service job is stressful (De Jonge and Dormann 2003), and it follows that job stress can hinder an employee's IRP. First, customer service employees typically believe that they lack control over their jobs and that they have the onerous task of satisfying conflicting demands of customers and managers. To magnify these issues, technology improvements have enabled service employees to handle more customer inquiries per hour, and employees rarely know the pace or nature of their work ahead of time (Schneider and Bowen 1995; Slepicka 2000). These factors contribute to job stress, detracting from the time that is necessary for such IRP tasks as record keeping and staying abreast of product knowledge. Second, job stress disrupts employees’ ability to perform prescribed tasks, because it also strips them of the energy resources that are necessary to perform (Hobfoll 2002). Thus, we expect job stress to negatively affect IRP.

We also expect job stress to affect CDERP. Because IRP requires formal reporting that is expected by management, some portion of a service employee's resources is likely devoted to IRP, leaving fewer resources to engage in CDERP. This notion is consistent with the COR/resource drain view, and job stress theorists note that an indication of elevated job stress is when employees stop engaging in CDERPs (Jex 1998). In addition, frontline employees may “disengage” from performing CDERPs, because job stress drains the types of emotional and cognitive resources (patience, empathy) that are necessary to offer stirring performances during the employee–customer interface (De Jonge and Dormann 2003; Kahn 1990).

H₂:

Job stress negatively affects (a) IRP and (b) CDERP.

Effects of performance on CPI

Perhaps no other customer outcome is a better indicator of loyalty than the intent to purchase again from a firm (Oliver 1997). Service employee performance may play a key role in influencing such intent. For example, service employee role-prescribed behavior has been found to affect customer satisfaction and loyalty (Bettencourt and Brown 1997). Furthermore, CDERPs, as rated by customers, have been found to positively affect customer perceptions of justice in a complaint-handling setting (Maxham and Netemeyer 2003). Conceptual writings also suggest that CDERP “delights” customers and enhances their intent to repurchase (Schneider and Bowen 1999).

H₃:

Both (a) IRP and (b) CDERP positively affect CPI.

Incremental Effects Model

FWC and WFC effects on CDERP

Although we expect that job stress is the primary predictor of IRP and CDERP, we also expect that FWC and WFC explain incremental variance in CDERP beyond that which is explained by job stress. Boundary spanners, such as customer service employees, must balance the demands of groups internal (supervisor, customers) and external (family) to the job. Compartmentalizing the cognitive and emotional resources that are necessary to fulfill the demands is difficult, negatively affecting certain aspects of performance, namely, CDERP. Furthermore, effective CDERP requires employees to engage customers, patiently listen to their concerns, and quickly develop suitable resolutions. Employees might not be able to perform these critical work roles effectively when they are expending considerable cognitive and emotional resources to resolve conflicts with their lives outside of their jobs (Kahn 1990). As such, employees may disengage from performing, and we posit that FWC negatively (and incrementally) affects CDERP.

We also contend that WFC has an incremental direct effect on CDERP. Edwards and Rothbard (2000) suggest that conflict in one domain reduces the personal capability to meet demands in the other domain, thereby inhibiting role performance in the other domain. They suggest that such an effect is direct, as well as indirect, through a mediating mechanism such as mental/physical health (i.e., job stress). Identity theory further contends that competing salient roles (family and work) potentially have a spillover effect, implying that work that interferes with family may affect CDERP. Finally, in his conceptualization of being personally disengaged from work, Kahn (1990) suggests that conflict in the work–family interface directly affects performance, particularly in service-related jobs. He suggests that employees may withdraw from a work role because resources that are necessary for that role (i.e., patience, understanding, and cognitive effort endemic to CDERP) are the same resources that are devoted to an outside role (family).

H₄:

Beyond the effect of job stress on CDERP, (a) FWC and (b) WFC negatively affect CDERP.

Effect of job stress on CPI

We posit that beyond the effects of employee performances, job stress incrementally affects CPI. Our rationale is as follows: First, although employees are believed to self-regulate by trying to conserve resources that are necessary to perform their jobs, the resources are quickly depleted when the job situation is stressful (Muraven, Tice, and Baumeister 1998). In customer service settings, there is a heightened potential for contentiousness between service employees and customers, as customers impose a variety of cognitive, emotional, and behavioral demands on service employees that may lower overall customer evaluations (De Jonge and Dormann 2003).

Second, customers are likely to detect the manifestations of job stress in the forms of poor decision making, lack of self-control, carelessness, or inhospitable looks. In effect, this lowers the ability to generate CPI, because service employees who experience job stress may not provide quality interactions with customers. Such a process concurs with Kahn's (1990) view of disengagement in critical moments of the work role. Specifically, job stress may directly affect CPI by inhibiting the service employee from offering a stirring role performance when interacting with customers. Customers may sense this and give low CPI ratings. Finally, there are anecdotal writings that suggest that when employees experience less job stress, customers rate the service they receive more favorably, creating positive customer evaluations (Furnham 2002; Schneider and Bowen 1999).

H₅:

Beyond the positive effects of the performance constructs on CPI, job stress has an incremental negative effect on CPI.

Relative Effects of WFC, FWC, Job Stress, and Performance

WFC versus FWC and their effects on CDERP

It is important to understand whether FWC or WFC has a more detrimental effect on CDERP. Conventional wisdom and interrole conflict theory suggest that FWC should have a greater effect. That is, FWC reflects the degree to which the family role interferes with performing work-related tasks, whereas WFC reflects the degree to which the job role interferes with performing family-related tasks. Interrole conflict theory further suggests that negative behavioral outcomes are more likely experienced in the domain to which the source of the conflict is directed (i.e., family interfering with work) (Netemeyer, Boles, and McMurrian 1996). Finally, the employee resources that are necessary to fulfill the family role (patience, understanding, and empathy) are the same resources that are endemic to CDERP. As the family role depletes the resources, less of the resource is available for CDERP. Thus, we expect that FWC has a stronger effect on CDERP than does WFC.

H₆:

The effect of FWC on CDERP is stronger than the effect of WFC on CDERP.

Job stress and its effect on IRP versus CDERP

Does job stress have a stronger effect on IRP or on CDERP? Considering the type of extra-role performance that we examine, we believe that job stress has a stronger effect on IRP. Identity theory suggests that the strength of identification with a role dictates the resources that are applied to the role. This tenet also holds for a role within a given role (IRP versus CDERP within the overall work role) (Rothbard and Edwards 2003). Specifically, the goals of the firm are at the core of an employee's work-role investment. In a customer service context, a firm's goal is to maximize customer loyalty, and employees may allow job stress to affect IRP more than CDERP because they identify more strongly with the CDERP aspect of the work. In addition, consistent with the attraction–selection–attrition framework (Schneider, Goldstein, and Smith 1995) and Bitner, Booms, and Mohr's (1994) work, boundary spanners are attracted to their positions because they possess the characteristics and the desire to provide good service. Finally, in our samples, the service employees went through extensive selection, training, and socialization processes. These processes emphasized the goal of customer retention by “going the extra mile” and “taking ownership” of a customer's problem. On the basis of this rationale, we expect that IRP suffers more at the hands of job stress than does CDERP.

H₇:

Job stress has a stronger effect on IRP than on CDERP.

CDERP versus IRP and their effects on CDERP

We posit that CDERP has a stronger effect on CPI than on IRP for three reasons. First, customers are more likely to witness extra-role behaviors. Given that many IRPs are not seen or do not resonate with customers unless they are exceptionally good or poor, it seems reasonable that the effect of IRP is not as strong as that of CDERP. Furthermore, because extra-role behaviors go “beyond the call of duty,” these behaviors are more likely to strike a chord with customers and resonate with them over time. This notion is consistent with the writings of Bitner, Booms, and Tetreault (1990). In their analysis of critical customer service incidents, they report that “extraordinary” service employee behaviors, such as courtesy and thoughtfulness, translated into customers’ experiencing high levels of satisfaction and remembering the service encounter.

Second, customers tend to recall encounters that are vivid or distinct, such as those involved in our customer complaint settings (Price, Arnould, and Tierney 1995). Furthermore, given that CDERPs are discretionary, customers may view the behaviors as internal to the customer service employee with whom they dealt. Similar to the cognitive processes that underlie managers’ performance ratings of employees, customers may attribute CDERPs as stable and internal to the employee (i.e., “this person is giving great service because he or she wants to, not because he or she has to”). Such CDERPs are likely to be retained in memory and recalled in deriving final evaluations (i.e., CPI) (e.g., MacKenzie, Podsakoff, and Fetter 1993). Thus, CDERP should have a greater effect on CPI than does IRP.

Finally, service employees are hired to handle customer issues, and in managers’ communication to employees (as in our samples that follow), they reiterate that in-role job duties are primarily tools that may help employees solve customer problems. It is also strongly stressed to employees that they must “see the big picture” of the organizational goal of retaining customers by solving customer problems. It would seem that CDERP is more instrumental to that goal than is IRP.

H₈:

CDERP has a stronger effect on CPI than does IRP.

Study Methods

Sample 1: Procedures and Measures

Procedures

With our first sample, we gathered data from customers, customer service employees, and supervisors of an online electronics retailer (i.e., a business-to-consumer setting). Customers initiated contact with the firm by telephoning the customer service center, which routed them to a service employee. (The primary responsibility of each employee was to handle customer complaints.) The employee logged a “job problem report” into the customer database and then attempted to resolve the complaint. All customer queries appeared in a complaint register, which was closely monitored by service employee supervisors. After exhausting all efforts to resolve a complaint, customer service employees “closed out” the job problem report in the customer database and noted the event history in the buyer's profile.

Customer measures

Our data-collection process began with customers. A total of 700 surveys were e-mailed to customers after they experienced a customer service interaction. Of the 700 e-mailed surveys, 346 were returned, of which 320 had complete responses across all study variables, yielding a 45.7% customer response rate. Within the six-month study period, each of the 320 customers dealt exclusively with only one service employee, and in all cases, the complaint that the customers registered was their first ever with the firm.

After the customer's most recent service experience with the firm and its employee, customers rated three CPI items on a seven-point scale, ranging from “not likely” to “very likely.” Given that customer satisfaction is commonly viewed as an antecedent of CPI, we created a measure of satisfaction based on the complaint/service request we described previously. We used the measure as a control variable to estimate all models that follow. In total, 60% of the sample was male, 69% had been customers for two to four years, 51% reported incomes from $40,000 to $60,000, and 74% held college degrees.

Employee measures

After a customer service inquiry was closed out, an e-mail containing a survey was sent to the service employee who handled the complaint. Surveys were sent to 700 service employees who specifically handled the 700 customer-initiated complaints. Of the surveys sent to customer service employees, 621 usable responses were returned, yielding an 88.7% response rate. In total, 320 surveys were matched to the fully completed customer survey, yielding a 45.7% response rate. That is, 320 service employee surveys were matched to the specific customer survey that represented the customer whom they had served. Thus, all analyses involving Sample 1 use a sample of 320 customer–employee response pairs.

The survey asked employees about their perceptions of the focal constructs (i.e., WFC, FWC, and job stress) over the past six months. We measured WFC and FWC with three items from the scales that Netemeyer, Boles, and McMurrian (1996) developed. We measured job stress with three items from the anxiety–stress scale that House and Rizzo (1972) developed and with one item that we generated. We also gathered two other measures from service employees. Job satisfaction and affective organizational commitment (OC) have been shown to be potential outcomes of WFC, FWC, and job stress and to be potential antecedents and/or outcomes of IRP and CDERP (MacKenzie, Podsakoff, and Ahearne 1998; Podsakoff et al. 2000). We used a three-item measure of job satisfaction and a four-item measure of OC (Mowday, Steers, and Porter 1979), and we used these measures as control variables to estimate all models that follow. All items were rated on a seven-point scale, ranging from “strongly disagree” to “strongly agree.” All employees were full time, 53% were female, 82% had been employees for five years or fewer, 81% reported incomes from $40,000 to $70,000, and 67% held college degrees.

Supervisor measures

The 53 supervisors who managed the 320 employees were also sent e-mailed surveys, and all 53 participated. The supervisors managed an average of six service employees and rated their performance for the six-month study period. The supervisor surveys were returned and were matched to the respective employees whom they rated. Given that supervisors rated more than one employee, we created a partially nested design between employees and supervisors. In addition to estimating our structural equation models (see Figure 1), we estimated a set of linear mixed models that control for any random effects of the supervisors’ repeated measurements on the parameter estimates that involve the supervisor–employee pairs (H₂–H₃). For both samples, there was no significant across- or within-supervisor rating biases due to evaluating multiple employees within a workgroup.

We measured employees’ IRP with three items that we adapted from Singh Verbeke, and Rhoads's (1996) work; the items were rated on a seven-point scale, ranging from “among the worst in the company” to “among the best in the company.” We measured CDERP with four items that we adapted from Bettencourt and Brown's (1997) work; the items were rated on a seven-point scale, ranging from “never” to “as often as possible.” We used supervisor-rated measures of employee performance for several reasons. First, employees may overrate their performance (Murphy and Cleveland 1995). Second, evidence suggests that as predictors of a criterion variable other than that rated by employees or supervisors, supervisor-rated measures of employee performance have the stronger predictive validity (Atkins and Wood 2002). Other scholars also suggest that supervisor-rated measures of employee performance are more valid than are employee self-ratings (Scullen, Mount, and Goff 2000). Finally, we used a supervisor-rated measure of employee performance because customers rated the prime dependent variable of interest, CPI. Thus, the links between the performance constructs and CPI are free of same-source bias (Podsakoff et al. 2003). Of the supervisors, 60% were female, 80% had been with the firm for eight years or more, 77% reported incomes from $55,000 to $70,000, and 57% held college degrees (the Appendix shows the measures for the focal constructs of our studies [i.e., CPI, WFC, FWC, job stress, IRP, and CDERP]).¹

Three measurement issues are of note. First, we began with 23 employee-rated items. Using confirmatory factor analyses and our own judgment, we trimmed this pool to 17 items (3 WFC, 3 FWC, 4 job stress, 3 job satisfaction, and 4 OC). We began with 9 supervisor- and 6 customer-rated items. Using confirmatory factor analyses, we retained 7 supervisor items and 6 customer items. Second, because we are linking performance measures that span a six-month period to a specific customer's purchase intents, we assume that performance is stable over time. There is some evidence to support this assumption; Viswesvaran, Ones, and Schmidt (1996) report a meta-analytic correlation of .81 for supervisor ratings of employee performance over two time periods. The strength of the correlation is indicative of stable performance over time. Third, although the performance–CPI links are free of same-source bias, WFC, FWC, and job stress items were rated by the same source. We compared the path estimates of a same-source methods factor model to the path estimates from a model without the same-source factor (Williams and Anderson 1994). We found that the path estimates were somewhat attenuated when the same-source factor was present, but the path estimates were not significantly affected.

Sample 2: Procedures and Measures

Procedures

Our second sample used 132 customers, service employees, and supervisors of a manufacturer/seller of technology-related equipment for retailers and financial service institutions (i.e., business-to-business setting). Customers initiated a call to the service employee, who personally visited the customer's site and tried to resolve a problem that the customer was having with equipment purchased from the manufacturer. The majority of complaints involved software/hardware malfunctions, and supervisors often accompanied service employees. We used the exact same data-collection procedures as we did for Sample 1.

Customer measures

Surveys were e-mailed to customers after they experienced a customer service interaction. Of the 200 surveys that were sent, 132 were returned with complete responses across the same customer measures that we used in Sample 1 (a 66% customer response rate). The 132 customer respondents dealt specifically and exclusively with one of the 132 service employees over the six-month study period. These customers were responsible for the purchasing and day-to-day functioning of the technical equipment used in their firms. In total, 67% of the customers were female, and 82% held college degrees.

Employee measures

After the customer service inquiries were resolved, a survey was e-mailed to the 200 service employees who specifically handled the 200 customer-initiated complaints. Of the surveys sent, 185 responses were returned, yielding a 92.5% employee response rate. Of these, 132 surveys were matched to the completed customer survey, yielding a 66% matched response rate. Thus, all analyses for Sample 2 use a sample of 132 customer–employee response pairs. The employees completed the same measures that we reported in Sample 1. All employees were full time, 67% were female, 82% had been employees for three years or fewer, 72% reported incomes from $50,000 to $70,000, and 87% held college degrees.

Supervisor measures

The 27 supervisors who managed the 132 employees were also sent e-mailed surveys. Each supervisor managed an average of five employees and rated the employees for the six-month study period over the same performance measures that we used in Sample 1. The supervisor surveys were matched to the respective service employee survey. In total, 42% of the supervisors were female, 60% had been employees for six years or more, 78% reported incomes from $50,000 to $70,000, and 65% held college degrees.²

Measurement Checks

Discriminant validity and reliability

We examined the psychometric properties of our measures by estimating a 30-item, 9-factor (WFC, FWC, job stress, job satisfaction, OC, IRP, CDERP, CPI, and customer satisfaction) measurement model. This model fit the data well for Sample 1 (χ² = 882.35, degrees of freedom [d.f.] = 369; comparative fit index [CFI] = .94; nonnormed fit index [NNFI] = .93; and root mean square error of approximation [RMSEA] = .06) and for Sample 2 (χ² = 592.29, d.f. = 369; CFI = .93; NNFI = .92; and RMSEA = .06) (Hu and Bentler 1995). Tables 1 and 2 show summary statistics, internal consistency estimates, and correlations for Samples 1 and 2, respectively. Coefficient alpha estimates of internal consistency ranged from .61 to .95, and average variance estimates (AVE) ranged from .30 to .87. The disattenuated correlations among the constructs (ϕ estimates) ranged from −.79 (job stress and IRP) to .72 (IRP and CDERP). We examined discriminant validity among the constructs by comparing ϕ to AVE. If the square of the parameter estimate between two constructs (ϕ²) is less than the average AVE between the two constructs, discriminant validity is supported (Fornell and Larcker 1981). This criterion was met for all possible pairs of constructs in Tables 1 and 2.

Table 1

Summary Statistics, Internal Consistency Estimates, and Correlations Among Constructs: Sample 1 (N = 320)

	Mean	Standard Deviation	α	AVE	1	2	3	4	5	6	7	8	9
Focal Constructs
1. WFC	5.86	1.17	.91	.78	1.00
2. FWC	5.44	.82	.84	.70	.69	1.00
3. Job Stress	5.47	.86	.89	.73	.64	.41	1.00
4. IRP	3.83	1.39	.94	.85	−.36	−.28	−.79	1.00
5. CDERP	3.01	1.22	.95	.87	−.69	−.52	−.68	.72	1.00
6. CPI	3.24	1.03	.90	.75	−.75	−.47	−.76	.63	.62	1.00
Control Variables
7. Job satisfaction	4.69	.88	.77	.60	−.32	−.17	−.52	.42	.24	.54	1.00
8. OC	4.44	.60	.63	.31	−.29	−.27	−.51	.50	.27	.61	.52	1.00
9. Customer satisfaction	5.48	1.00	.83	.65	−.51	−.45	−.52	.55	.61	.40	.17	.19	1.00

Notes: Means and standard deviations are based on summated scale averages. Correlations are disattenuated estimates from a confirmatory-factor measurement model.

Table 2

Summary Statistics, Internal Consistency Estimates, and Correlations Among Constructs: Sample 2 (N = 132)

	Mean	Standard Deviation	α	AVE	1	2	3	4	5	6	7	8	9
Focal Constructs
1. WFC	5.81	1.22	.94	.84	1.00
2. FWC	5.36	.85	.80	.64	.63	1.00
3. Job Stress	5.55	.84	.87	.66	.60	.46	1.00
4. IRP	3.76	1.39	.93	.83	−.37	−.36	−.76	1.00
5. CDERP	3.11	1.22	.92	.82	−.73	−.60	−.65	.69	1.00
6. CPI	3.25	1.05	.89	.72	−.72	−.50	−.73	.66	.66	1.00
Control Variables
7. Job satisfaction	4.65	.90	.75	.59	−.29	−.22	−.52	.38	.27	.53	1.00
8. OC	4.50	.57	.61	.30	−.24	−.33	−.43	.49	.27	.66	.53	1.00
9. Customer satisfaction	5.50	1.00	.87	.71	−.47	−.45	−.50	.51	.58	.40	.15	.10	1.00

Notes: Means and standard deviations are based on summated scale averages. Correlations are disattenuated estimates from a confirmatory-factor measurement model.

We conducted several sample bias checks. For both samples, we compared the demographic characteristics of our employee, supervisor, and customer samples with several other samples that did not participate in our study. We found no significant differences between our samples and the other samples across demographics. We also assessed potential effects of employee gender, age, and education. As predictors or moderators of WFC and FWC, these variables had no effects. For Sample 2 (n = 132), we assessed whether marital status and having child dependents affected the results. These variables had no effects as either predictors or moderators of WFC and FWC.

Multigroup measurement invariance

We also tested whether the measures were invariant across samples by estimating a hierarchy of multigroup measurement invariance models (Steenkamp and Baumgartner 1998). The first model that we estimated was a baseline measurement model of the same pattern of fixed and free parameters. This model fit the data well (χ² = 1474.64, d.f. = 738; CFI = .94; NNFI = .93; and RMSEA = .06). Next, we estimated a model that constrained the factor loadings to be invariant across groups. The fit indices for this model were acceptable (χ² = 1491.11, d.f. = 759; CFI = .94; NNFI = .93; and RMSEA = .06), and the difference in fit between this model and the baseline measurement model was not significant (χ²_diff = 16.47, d.f._diff = 21, p > 10). Thus, the loadings for all measurement items were invariant across samples. The third model that we estimated specified invariant factor variances and covariances in addition to invariant factor loadings. It had acceptable fit indices (χ² = 1510.29, d.f. = 804; CFI = .94; NNFI = .93; and RMSEA = .06), and the difference in fit between this model and the baseline measurement model was not significant (χ²_diff = 35.65, d.f._diff = 66, p > 10). Thus, all correlational (covariance) relations among model constructs, as well as construct (factor) variances, were equal across samples. Finally, we estimated a fully invariant measurement model (i.e., invariant factor loadings, invariant factor variances and covariances, and invariant item error loadings). Although the indices showed adequate fit (χ² = 1659.54, d.f. = 834; CFI = .93; NNFI = .93; and RMSEA = .07), this model's fit differed from the fit of the baseline measurement model (χ²_diff = 184.90, d.f._diff = 96, p < .01). Still, the invariance of the factor loadings, factor variances, and covariances suggests that comparing model paths across samples is appropriate.

Analyses and Results

Structural Model Procedures and Results

To test the structural models, we used a multiple-group approach akin to the method that Singh (2000) uses. With this approach, we estimated two multigroup models. The first model freely estimates all paths across samples (i.e., the baseline structural model). The model is then compared with a model in which all paths are constrained to be equal across samples (i.e., a full structural invariance model). If the difference in fit between the two models is not significant, there is evidence that all structural paths are invariant across samples. If structural invariance exists, hypothesized paths can be tested with the multigroup approach.

Mediated model

We first used this approach for the mediated model that appears in Figure 1. As we previously noted, we measured three control variables: service employee job satisfaction, OC, and customer satisfaction. As is the recent trend in structural equation models to control for variables that are not central to study hypotheses (Baltes and Heydens-Gahir 2003; Rothbard and Edwards 2003), we also estimated paths among the control variables and the focal constructs (prior research suggests 14 paths among the focal constructs and control variables; Podsakoff et al. 2000; Oliver 1997). Thus, for all structural models that follow, the paths among the focal constructs (“Hypothesized Paths” in Table 3) reflect the effects of simultaneously estimating paths among the focal constructs and control variables (“Control Variable Paths” in Table 3).

Table 3

Multigroup Full Structural Invariance Model Results

	A: Model Fit
	χ²	d.f	χ²_diff(3)	CFI	NNFI	RMSEA
Mediated model	2089.83	788	—	.89	.88	.08
Incremental effects model	1884.80	785	205.03	.91	.90	.07

	B: Path Estimates
	Mediated Model				Incremental Effects Model
Hypothesized Paths	Unstandardized	Standardized	t-Value	Supported	Unstandardized	Standardized	t-Value	Supported
H_1a: WFC → job stress	.51	.62	10.44	Yes	.55	.62	10.45	Yes
H_1b: FWC → job stress	.05	.04	.79	No	.02	.02	.33	No
H_2a: Job stress → IRP	–1.15	−.80	19.30	Yes	–1.14	−.80	19.14	Yes
H_2b: Job stress → CDERP	–1.10	−.87	15.38	Yes	−.72	−.56	9.19	Yes
H_3a: IRP → CPI	.32	.42	7.85	Yes	−.05	−.06	.92	No
H_3b: CDERP → CPI	.33	.38	6.63	Yes	.18	.21	3.73	Yes
H_4a: FWC → CDERP	—	—			−.18	−.14	2.94	Yes
H_4b: WFC → CDERP	—	—			−.34	−.32	5.73	Yes
H₅: Job stress → CPI	—	—			−.82	−.74	9.02	Yes
R²: Job stress	.45				.44
R²: IRP	.64				.65
R²: CDERP	.57				.69
R²: CPI	.53				.69
Control Variable Paths
WFC → job satisfaction	.05	.05	.61		.08	.08	1.12
WFC → OC	.09	.33	2.48		.08	.21	2.32
FWC → job satisfaction	.08	.07	1.08		.05	.04	.60
FWC → OC	−.09	−.18	2.36		−.10	−.20	2.61
IRP → job satisfaction	−.06	−.07	.79		−.13	−.16	1.90
Job satisfaction → OC	.13	.32	4.09		.12	.30	3.85
Job satisfaction → CDERP	−.16	−.15	3.13		−.15	−.14	3.14
OC → CDERP	−.27	−.10	1.68		−.25	−.09	1.60
Job stress → job satisfaction	−.78	−.64	6.40		−.90	−.74	7.57
Job stress → OC	−.21	−.44	4.35		−.22	−.45	4.40
IRP → customer satisfaction	.11	.16	1.99		.11	.16	2.06
CDERP → customer satisfaction	.32	.42	6.08		.33	.43	6.53
Job stress → customer satisfaction	−.10	−.11	1.09		−.09	−.10	1.02
Customer satisfaction → CPI	−.06	−.05	.98		−.10	−.09	1.89

The baseline structural model of freely estimated paths across samples demonstrated fit levels approaching adequacy (χ² = 2082.60, d.f. = 768; CFI = .89; NNFI = .88; and RMSEA = .08), as did the full structural invariance model (χ² = 2089.83, d.f. = 788; CFI = .89; NNFI = .88; and RMSEA = .08). Given that the difference in fit between the two models was not significant (χ²_diff = 7.23, d.f._diff = 20, p > .10), all paths are equal across groups, and we offer the results for the mediated model based on full structural invariance.

We predicted that WFC and FWC would affect (produce) job stress (H₁). As we show in Table 3, the WFC → job stress path was significant and positive, in support of H_1a. Although correlated with job stress (ϕ = .41 and .46 from Tables 1 and 2), the FWC → job stress path was not significant, likely because of the shared variance between WFC and FWC. Still, the WFC and FWC paths explained 45% of the variance in job stress. H₂ posits direct effects of job stress on IRP and CDERP. As we show in Table 3, both hypotheses were supported, explaining large portions of the variance in IRP (R² = .64) and CDERP (R² = .57). Finally, H₃ of the mediated model predicts that both IRP and CDERP positively affect CPI. Both paths were supported, explaining 53% of the variance in CPI.

Incremental effects model

H₄ posits that both FWC and WFC incrementally affect CDERP (i.e., job stress does not fully mediate the WFC/FWC effects on CDERP). H₅ posits a direct effect of job stress on CPI—that is, job stress explains incremental variance in CPI beyond that which IRP and CDERP explain. To test the hypotheses, we added paths from FWC and WFC to CDERP and a path from job stress, to CPI, to the mediated model to create the incremental effects model. Again, we used a multigroup model approach in which the baseline structural model of freely estimated paths across samples showed adequate fit (χ² = 1875.85, d.f. = 762; CFI = .90; NNFI = .89; and RMSEA = .07). The full structural invariance model also fit well (χ² = 1884.80, d.f. = 785; CFI = .91; NNFI = .90; and RMSEA = .07), and its difference in fit from the baseline structural model was not significant (χ²_diff = 8.95, d.f._diff = 23, p > .10), again suggesting that all paths are equal across groups. Thus, in Table 3, we offer the results for the incremental effects model based on full structural invariance.

Consistent with the writings on mediation (Iacobucci and Duhachek 2003), we first compared the incremental effects model to the mediated model. The difference in fit between the two models was significant (χ²_diff = 205.03, d.f._diff = 3, p < .01), suggesting that job stress does not totally mediate the effects of FWC and WFC on CDERP and that IRP and CDERP do not totally mediate the effect of job stress on CPI. The path estimates in Table 3 reinforce this conclusion. The FWC → CDERP and WFC → CDERP paths were significant, and they explained an additional 12% of the variance in CDERP beyond that explained by job stress in the mediated model. The job stress → CPI path was also strong, and it explained an additional 16% of the variance in CPI beyond that explained by IRP and CDERP in the mediated model.

Relative Effects of WFC, FWC, Job Stress, and Performance

By employing multigroup analyses with the full structural invariance incremental effects model, we tested our relative effects hypotheses. We tested each hypothesis with a chi-square difference test at 1 d.f. in which the two paths are constrained to be equal in one model but are freely estimated in another. We show the results in Table 4. Although H₆ posits that the direct effect of FWC on CDERP is stronger than the direct effect of WFC, the hypothesis was not supported. Both FWC and WFC equally affected CDERP. We tested H₇ by comparing the job stress → IRP path to the job stress → CDERP path, and we found that job stress had a greater effect on IRP than on CDERP. Finally, H₈ was supported because CDERP had a stronger direct effect on CPI than did IRP.

Table 4

Relative Effects of WFC, FWC, Job Stress, and Performance

Path Comparison Hypotheses	χ²_diff	Supported
H₆: FWC → CDERP > WFC → CDERP	2.15 (n.s.)	No
H₇: Job stress → IRP > job stress → CDERP	18.28	Yes
H₈: CDERP → CPI > IRP → CPI	8.46	Yes

Notes: We conducted all χ²_diff tests at a difference of 1 d.f. by comparing the model that constrained the paths of interest to be equal to the model in which the paths were freely estimated—that is, the incremental effects model with a χ² of 1884.80 and d.f. of 785. Except where noted as “n.s.” (not significant), all χ² differences were significant at the .05 level or better.

Indirect and Total Effects of WFC and FWC

As we noted previously, WFC had strong direct effects on job stress and CDERP, and FWC had a direct effect on CDERP. Although not hypothesized, WFC and FWC showed a series of important indirect and total effects as well. Indirect effects are those that are mediated by at least one other variable, and total effects are the sum of the direct (e.g., WFC → CDERP) and indirect (e.g., WFC → job stress → CDERP) effects (Bollen 1989). The indirect effect of WFC on CDERP through job stress was significant (unstandardized indirect path = −.31, standardized indirect path = −.29; t = 6.97, p < .01), as was the total effect of WFC on CDERP (unstandardized total effect = −.65, standardized total effect = −.69; t = 11.4l, p < .01). The indirect effects of WFC on IRP (unstandardized indirect path = −.59, standardized indirect path = −.49; t = 9.67, p < .01) and CPI (unstandardized indirect path = −.48, standardized indirect path = −.51; t = 10.18, p < .01) were also significant. Although FWC did not show a direct effect on job stress, it showed a direct effect on CDERP and had one total effect. The FWC total effect on CDERP was significant (unstandardized total effect = −.18, standardized total effect = −.13; t = 2.62, p < .01).

Discussion

Summary of Key Findings

Effects of WFC and FWC

Both FWC and WFC had negative incremental effects on CDERP. Thus, their effects on CDERP were not totally mediated by job stress. Furthermore, WFC showed strong indirect effects on IRP, CDERP, and CPI.

Effects of job stress and performance

Job stress affected both aspects of performance, and job stress was more strongly related to IRP than to CDERP. Both IRP and CDERP affected CPI in the mediated model. However, job stress had a strong negative incremental effect on CPI that was not totally mediated by the positive effects of performance. Finally, CDERP had a stronger effect on CPI than did IRP.

Managerial Implications

Our results have important managerial implications. However, to be beneficial, all the initiatives and programs that we subsequently discuss require a real cultural shift in the way an organization views its employees. First, managers should be trained to regard their employees as family members, with all the issues and responsibilities that go with it, and they must understand that service employees do not leave family problems at the workplace doorstep (Rifkin 1996). Thus, just as service employees are asked to “take ownership” of customer problems, supervisors and managers may need to take at least some ownership of employee problems that stem from the work–family interface. Managers who take the attitude that “it's not my problem” may effectively be hurting the organizational goal of customer retention. Although training programs to teach managers these skills may seem like an obvious solution, many firms do not implement them because of the costs involved. One study reports that fewer than half (43%) of firms currently train managers to respond to the work– family needs of employees (Galinsky and Bond 1998). Our results provide some evidence that not training managers to deal with service employee job stress and its WFC and FWC antecedents may also carry a high cost.

Second, managers can structure the workplace so that employees are encouraged to balance work and family. Managers can implement family-friendly programs that send signals that management supports family roles. Some popular programs that members of Fortune‘s “100 Best Companies to Work For” have implemented include mentoring, child care, family leave, flextime, maternity/paternity leave, telecommuting, and on-site job counseling for work–family matters (Tkaczyk et al. 2003). Research shows that such programs can reduce work and family conflicts (Lobel 1999), enhance employee performance (Konrad and Mangel 2000), and positively correlate with a firm's share price (Arthur 2003). However, such programs are not a panacea in addressing WFC, FWC, and job stress effects. Management must be careful to design work–life initiatives with the specific needs of employees in mind. For example, work–life initiatives have been shown to be more effective for companies with an educated and professional workforce (such as our samples) (Konrad and Mangel 2000). Therefore, we recommend that managers assess sources of WFC, FWC, and job stress and design programs that include the most appropriate elements for their workforce. Not to consider each workforce individually may result in such policies being ineffective (Rau and Hyland 2002).

Third, managers can offer stress management workshops that, instead of offering employees advice on how to live life, ask for help in understanding how family roles affect work and what management can do about it (Osland, Kolb, and Rubin 2001). Workshops can create a welcoming forum in which employees freely provide suggestions for the better support of family roles. There is some evidence that workshops can reduce stress. Kossek, Colquitt, and Noe (2001) show that employee anxiety, irritability, and depression are reduced when managers encourage employees to openly discuss family concerns with supervisors and peers. Thus, even the simple notion of seeking employee feedback can help employees feel more valued and less stressed.

Fourth, properly trained managers can institute changes to the work itself. For example, employee compensation packages could be restructured to reward customer satisfaction in addition to rewarding prescribed in-role outcomes (e.g., call-time quotas). Indeed, some call centers have lowered expectations for the number of customer problems handled per shift and instead have empowered employees to take as long as is necessary to satisfy customer needs (Molpus 2003). Furthermore, managers could offer service employees a “voice” in determining aspects of their jobs. For example, managers could restructure systems for scheduled employee breaks (in their busy complaint-handling days) that allow employees to participate in the decision as to when the breaks are needed. Thus, managers and supervisors can create a culture that empowers employees, which may have the effect of helping employees balance the work–family interface and make the work itself less stressful.

Fifth, firms may benefit from bolstering the mental health benefits for employees who cope with stress and conflict that stem from the work–family interface. An initiative that seems particularly relevant is to assist employees in role and boundary management. Qualified mental health professionals can help employees manage personal relationships both at work and at home. There is evidence that the relationship between stressor and stress is largely the outcome of a person believing that he or she is unable to cope adequately with an identified threat (i.e., dissatisfied family or superiors) (Lazarus and Folkman 1984). Thus, firms should consider assisting employees in developing relationship skills to successfully juggle competing roles.

Theoretical Implications, Further Research, and Limitations

Incremental effects of WFC, FWC, and job stress

Traditional models view the effects of WFC and FWC (job stressors) on performance as likely being mediated by job stress and the effects of job stress on customer outcomes as likely being mediated by performance. However, our results suggest that this is not the case. In the high-stress customer service job, outside-role conflicts can directly affect performance, and within-domain stress can directly affect customer outcomes. Thus, future models may want to consider the context of the job when specifying relationships among WFC, FWC, job stress, performance, and customer outcomes. When conflicts in the work–family interface and job stress are high, these constructs may directly affect important behavioral outcomes for employees and customers.

Relative effects of WFC, FWC, and job stress

Both theory and intuition suggest that FWC has a stronger effect on CDERP than does WFC. However, the overall pattern of results suggests that this is not the case. Our results show a direct “spillover effect” of WFC that affects CDERP. Moreover, our results show that indirect and total effects of WFC on CDERP and CPI are stronger than are the indirect and total effects of FWC on CDERP and CPI. We offer two speculative explanations as to why we obtained such results.

First, perhaps employees are cognizant of the potential for FWC to affect their job performance, and thus they engage in processes or behaviors that partially compensate for the effect. However, the opposite may not be true. Employees may not be aware of the potential for WFC to affect job performance, and thus they may not engage in cognitive processes or behaviors that could reduce the effect. Second, it has recently been demonstrated that FWC is likely to have a strong effect on work performance when the pressure to participate in the family domain (pressure from the family) is strong and the pressure to participate in the work domain (pressure from the firm or supervisors) is concurrently weak (Greenhaus and Powell 2003). However, in customer service settings, the opposite may be more likely. Given the importance of employee performances in affecting customer evaluations, the pressure to participate in the work role could be stronger than the pressure to participate in the family role, resulting in a weaker FWC → CDERP path. Further studies could examine why the WFC → job performance spillover effect is so strong. What are other potential reasons for it? Is it as likely in less stressful marketing jobs?

With respect to the relative effects of job stress, we found that it affected IRP more than it did CDERP. In stressful customer service job situations, when employees believe that they are unable to perform all roles, they may perceive doing extra things that help affect customer evaluations as more salient to their job-role identity. Further studies might consider investigating different dimensions of performance and the relative effects of job stress on each dimension.

Limitations and further research

Our results are tempered with certain limitations. First, given that we collected WFC, FWC, and job stress measures cross-sectionally, establishing a temporal ordering among these constructs is not possible. Although our posited flow of WFC and FWC as correlated constructs that affect job stress is consistent with most research, other specifications are tenable (Williams and Alliger 1994). Although we estimated such alternative models and found a pattern of results similar to that which we report in Table 3, further studies that use experiments may better detect the causality among WFC, FWC, and job stress. For example, Greenhaus and Powell (2003) manipulated family- and work-role pressures to determine their effects on the decision to participate in a work activity versus a family activity. Such experiments may be useful to better show the causal flow among WFC, FWC, and job stress.

Second, we did not collect two critical measures of role salience: (1) the importance of the family role versus the job role for employees and (2) the importance of CDERP versus IRP. Such measures could have helped solidify the findings and bolster the use of identity theory as a basis for some of the hypotheses. Third, the WFC measure was more strongly related to job stress than was the FWC measure. Although this finding is consistent with the theoretical rationale for the WFC and FWC relationships with job stress, wording of the WFC items may suggest more of a content similarity to job stress than does the wording of the FWC items. Further studies might elect to use a job stress measure that assesses symptomatic health aspects related to work (e.g., high blood pressure, loss of appetite). Fourth, we examined only one customer per service employee, and we related customer-specific purchase intent to employee performance measures, encompassing a six-month period. Although extensive procedures suggest that our customer samples were not biased and the literature suggests that employee performance is relatively stable over time, further studies that examine multiple customers per employee are warranted.

Footnotes

Appendix Focal Construct Measures

References

Adams

Gary A.

, King

Lynda A.

, and King

Daniel W.

(1996), “Relationships of Job and Family Involvement, Family Social Support, and Work–Family Conflict with Job and Life Satisfaction,” Journal of Applied Psychology, 81 (4), 411–20.

Armour

Stephanie

(2002), “Workers Put Family First Despite Slow Economy, Jobless Fears,” USA Today, (June 6), 03B.

Arthur

Michelle M.

(2003), “Share Price Reactions to Work–Family Initiatives: An Institutional Perspective,” Academy of Management Journal, 46 (4), 497–505.

Atkins

P.W.B.

, and Wood

R.E.

(2002), “Self- Versus Other's Ratings of Assessment Center Ratings: Validation Evidence from 360-Degree Feedback Programs,” Personnel Psychology, 55 (4), 871–904.

Baltes

Boris B.

, and Heydens-Gahir

Heather A.

(2003), “Reduction of Work–Family Conflict Through the Use of Selection, Optimization, and Compensation Behaviors,” Journal of Applied Psychology, 88 (6), 1005–1018.

Beehr

Terry

(2000), “An Organizational Psychology Meta-Model of Occupational Stress,” in Theories of Organizational Stress, Cooper

C.L.

, ed. New York: Oxford University Press, 6–27.

Bettencourt

Lance A.

, and Brown

Stephen

(1997), “Contact Employees: Relationships Among Workplace Fairness, Job Satisfaction, and Pro-Social Service Behaviors,” Journal of Retailing, 73 (1), 39–61.

Bettencourt

Lance A.

, and Brown

Stephen

(2003), “Role Stressors and Customer-Oriented Boundary-Spanning Behaviors in Service Organizations,” Journal of the Academy of Marketing Science, 31 (4), 394–408.

Bettencourt

Lance A.

, Gwinner

Kevin P.

, and Meuter

Matthew L.

(2001), “A Comparison of Attitude, Personality, and Knowledge Predictors of Service-Oriented Organizational Citizenship Behaviors,” Journal of Applied Psychology, 86 (1), 29–41.

10.

Bitner

Mary Jo

, Booms

Bernard H.

, and Mohr

Lois A.

(1994), “Critical Service Encounters: The Employee's Viewpoint,” Journal of Marketing, 58 (October), 95–106.

11.

Bitner

Mary Jo

, Bernard

, and Tetreault

Mary Stanfield

(1990), “The Service Encounter: Diagnosing Favorable and Unfavorable Incidents,” Journal of Marketing, 54 (January), 71–84.

12.

Bollen

Kenneth A.

(1989), Structural Equations with Latent Variables. New York: John Wiley & Sons.

13.

Borman

Walter C.

, and Motowidlo

Stephan J.

(1993), “Expanding the Criterion Domain to Include Elements of Contextual Performance,” in Personnel Selection in Organizations, Schmitt

, and Borman

W.C.

, eds. San Francisco: Jossey-Bass, 71–98.

14.

Cox

Tom

, Griffiths

Amanda

, and Rial-Gonzalez

Eusebio

(2000), Research on Work-Related Stress. Luxemburg, Germany: Office for the Publications of the European Communities.

15.

De Jonge

Jan

, and Dormann

Christian

(2003), “The DISC Model Demand-Induced Strain Compensation Mechanisms in Job Stress,” in Occupational Stress in the Service Professions, Dollard

M.F.

, Winefield

A.H.

, and Winefield

H.R.

, eds. New York: Taylor & Francis, 43–74.

16.

Edwards

Jeffrey R.

, and Rothbard

Nancy P.

(2000), “Mechanisms Linking Work and Family: Clarifying the Relationship Between Work and Family Constructs,” Academy of Management Review, 25 (1), 178–99.

17.

Fornell

Claes

, and Larcker

David

(1981), “Evaluating Structural Equation Models with Unobservable Variables and Measurement Error,” Journal of Marketing Research, 18 (November), 39–50.

18.

Frone

Michael R.

(2000), “Work–Family Conflict and Employee Psychiatric Disorders: The National Comorbidity Study,” Journal of Applied Psychology, 85 (6), 888–95.

19.

Frone

Michael R.

, Russell

Marcia

, and Cooper

M. Lynne

(1992), “Antecedents and Outcomes of Work–Family Conflict: Testing a Model of the Work–Family Interface,” Journal of Applied Psychology, 77 (1), 65–78.

20.

Furnham

Adrian

(2002), “Happy Staff Is Not the Full Answer: Management Style Can Be Reflected in Customer Service— But the Relationship is Complex,” The Financial Times, (February 5), 16.

21.

Galinsky

Ellen

, and Bond

James T.

(1998), The 1998 Business Work–Life Study: A Source-Book, (Publication #W98–03). New York: Families and Work Institute, 1–26.

22.

Greenhaus

Jeffrey H.

, and Powell

Gary N.

(2003), “When Work and Family Collide: Deciding Between Competing Role Demands,” Organizational Behavior & Human Decision Processes, 90 (2), 291–303.

23.

Hobfoll

Stevan E.

(2002), “Social and Psychological Resources and Adaptation,” Review of General Psychology, 6 (4), 307–324.

24.

House

Robert J.

, and Rizzo

John R.

(1972), “Role Conflict and Ambiguity as Critical Variables in a Model of Organizational Behavior,” Organizational Behavior and Human Performance, 7 (3), 467–505.

25.

Li-tze

, and Bentler

Peter M.

(1995), “Evaluating Model Fit,” in Structural Equation Modeling: Concepts, Issues, and Applications, Hoyle

R.H.

, ed. Thousand Oaks, CA: Sage Publications, 76–99.

26.

Iacobucci

Dawn

, and Duhachek

Adam

(2003), “Mediation Analysis,” in Association for Consumer Research Round Table Presentations. Provo, UT: Association for Consumer Research, 1–20.

27.

Jex

Steve M.

(1998), Stress and Job Performance: Theory, Research, and Implications for Managerial Practice. Thousand Oaks, CA: Sage Publications.

28.

Kahn

Robert L.

, and Byosiere

Philippe

(1992), “Stress in Organizations,” in Handbook of Industrial and Organizational Psychology, Vol. 3, Dunnette

M.D.

, and Hough

L.M.

, eds. Palo Alto, CA: Consulting Psychologists Press,, 571–650.

29.

Kahn

William A.

(1990), “Psychological Conditions of Personal Engagement and Disengagement at Work,” Academy of Management Journal, 33 (4), 692–724.

30.

Konrad

Alison M.

, and Mangel

Robert

(2000), “The Impact of Work–Life Programs on Firm Productivity,” Strategic Management Journal, 21 (2), 1225–37.

31.

Kossek

Ellen

, Colquitt

E.J.A.

, and Noe

Raymond A.

(2001), “Care-Giving Decisions, Well-Being, and Performance: The Effects of Place and Provider as a Function of Dependent Type and Work–Family Climates,” Academy of Management Journal, 44 (1), 29–44.

32.

Lazarus

R. S.

, and Folkman

(1984), Stress, Appraisal, and Coping. New York: Springer.

33.

Lobel

S. A.

(1999), “Impacts of Diversity and Work–Life Initiatives in Organizations,” in Handbook of Gender and Work, Powell

G.N.

, ed. Thousand Oaks, CA: Sage Publications, 453–76.

34.

MacKenzie

Scott B.

, Podsakoff

Philip M.

, and Ahearne

Michael M.

(1998), “Some Possible Antecedents and Consequences of In-Role and Extra-Role Salesperson Performance,” Journal of Marketing, 62 (July), 87–98.

35.

MacKenzie

Scott B.

, Philip

, and Fetter

Richard

(1993), “The Impact of Organizational Citizenship Behavior on Evaluations of Salesperson Performance,” Journal of Marketing, 57 (January), 70–80.

36.

Maxham

James G.

, and Netemeyer

Richard G.

(2003), “Firms Reap What They Sow: The Effects of Employee Shared Values and Perceived Organizational Justice on Customer Evaluations of Complaint Handling,” Journal of Marketing, 67 (January), 46–62.

37.

Molpus

David

(2003), “Customers Frustrated with Customer Service,” National Public Radio, Morning Edition, (January 6).

38.

Motowidlo

Stephan J.

, and Scotter

James R. Van

(1994), “Evidence That Task Performance Should Be Distinguished from Contextual Performance,” Journal of Applied Psychology, 79 (4), 475–80.

39.

Mowday

Richard T.

, Steers

Richard M.

, and Porter

Lyman W.

(1979), “The Measurement of Organizational Commitment,” Journal of Vocational Behavior, 14 (2), 224–47.

40.

Muraven

Mark

, Tice

Dianne M.

, and Baumeister

Roy

(1998), “Self-Control as a Limited Resource: Regulatory Depletion Patterns,” Journal of Personality and Social Psychology, 74 (3), 774–89.

41.

Murphy

Kevin R.

, and Cleveland

Jeanette N.

(1995), Understanding Performance Appraisals: Social, Organizational, and Goal-Based Perspectives. Thousand Oaks, CA: Sage Publications.

42.

National Institute for Occupational Safety and Health (2002), Stress at Work, (Publication No. 99–101). Cincinnati: Department of Health and Human Services.

43.

Netemeyer

Richard G.

, Boles

James S.

, and McMurrian

Robert

(1996), “Development and Validation of Work–Family Conflict and Family–Work Conflict Scales,” Journal of Applied Psychology, 81 (4), 400–410.

44.

Oliver

Richard

(1997), Satisfaction: A Behavioral Perspective on the Consumer. Boston: Irwin/McGraw-Hill.

45.

Osland

Joyce S.

, Kolb

David A.

, and Rubin

Irwin M.

(2001), Organizational Behavior: An Experiential Approach, 7th ed. Upper Saddle River, NJ: Prentice Hall.

46.

Podsakoff

Philip M.

, MacKenzie

Scott B.

, Lee

Jeong-Yeon

, and Podsakoff

Nathan P.

(2003), “Common Methods Bias in Behavioral Research: A Critical Review of the Literature and Recommended Remedies,” Journal of Applied Psychology, 88 (5), 879–903.

47.

Podsakoff

Philip M.

, Scott

, Paine

Julie B.

, and Bachrach

Daniel G.

(2000), “Organizational Citizenship Behaviors: A Critical Review of the Theoretical and Empirical Literature and Suggestions for Future Research,” Journal of Management, 26 (3), 513–63.

48.

Price

Linda L.

, Arnould

Eric J.

, and Tierney

(1995), “Going to Extremes: Managing Service Encounters and Assessing Provider Performance,” Journal of Marketing, 59 (April), 83–97.

49.

Rau

Barbara L.

, and Hyland

Maryanne M.

(2002), “Role Conflict and Flexible Work Arrangements: The Effects on Applicant Attraction,” Personnel Psychology, 55 (Spring), 111–36.

50.

Rifkin

Glenn

(1996), “Finding Meaning at Work,” Strategy and Business, 1 (4th quarter), 15–17.

51.

Rothbard

Nancy P.

, and Edwards

Jeffrey R.

(2003), “Investment in Work and Family Roles: A Test of Identity and Utilitarian Motives,” Personnel Psychology, 56 (3), 699–730.

52.

Schneider

Benjamin

, and Bowen

David

(1995), Winning the Service Game. Boston: Harvard Business School Press.

53.

Schneider

Benjamin

, and Bowen

David

(1999), “Understanding Customer Delight and Outrage,” Sloan Management Review, 41 (1), 35–46.

54.

Schneider

Benjamin

, Goldstein

Harold W.

, and Smith

D. Brent

(1995), “The ASA Framework: An Update,” Personnel Psychology, 48 (Winter), 747–73.

55.

Scullen

Steven E.

, Mount

Michael K.

, and Goff

Maynard

(2000), “Understanding the Latent Structure of Job Performance Ratings,” Journal of Applied Psychology, 85 (6), 956–70.

56.

Singh

Jagdip

(2000), “Performance Productivity and Quality of Frontline Employees in Service Organizations,” Journal of Marketing, 64 (April), 15–34.

57.

Singh

Jagdip

, Verbeke

William

, and Rhoads

Gary K.

(1996), “Do Organizational Practices Matter in Role Stress Processes? A Study of Direct and Moderating Effects for Marketing-Oriented Boundary Spanners,” Journal of Marketing, 60 (July), 69–86.

58.

Slepicka

Mary

(2000), “Verizon Addresses CSR Stress Issues,” Customer Interface, (September), 11, 21.

59.

Steenkamp

J.E.M.

, and Baumgartner

Hans

(1998), “Assessing Measurement Invariance in Cross-National Consumer Research,” Journal of Consumer Research, 25 (June), 78–90.

60.

Thoits

Peggy A.

(1991), “On Merging Identity Theory and Stress Research,” Social Psychology Quarterly, 54 (2), 101–112.

61.

Thoits

Peggy A.

(1995), “Stress, Coping, and Social Support Processes: Where Are We? What Next?” Journal of Health and Social Behavior, (Extra Issue), 53–79.

62.

Tkaczyk

Christopher

, Harrington

Ann

, Moskowitz

Milton

, and Levering

Robert

(2003), “100 Best Companies To Work For,” Fortune, 147 (1), 127–40.

63.

Viswesvaran

Chockalingam

, Ones

Deniz S.

, and Schmidt

Frank L.

(1996), “Comparative Analysis of the Reliability of Job Performance Ratings,” Journal of Applied Psychology, 81 (5), 557–74.

64.

Williams

Kevin J.

, and Alliger

George M.

(1994), “Role Stressors, Mood Spillover, and Perceptions of Work–Family Conflict in Employed Parents,” Academy of Management Review, 37 (4), 837–68.

65.

Williams

Lawrence J.

, and Anderson

Stella E.

(1994), “An Alternative Approach to Methods Effects by Using Latent-Variable Models: Applications in Organizational Behavior Research,” Journal of Applied Psychology, 79 (2), 323–31.